Freeze Damage to Polymer Electrolyte Fuel Cells

Freeze Damage to Polymer Electrolyte Fuel Cells

Chapter 6 Freeze Damage to Polymer Electrolyte Fuel Cells Abdul-Kader Srouji1 and Matthew M. Mench2* 1 Fuel Cell Dynamics and Diagnostics Laboratory,...

3MB Sizes 1 Downloads 104 Views

Chapter 6

Freeze Damage to Polymer Electrolyte Fuel Cells Abdul-Kader Srouji1 and Matthew M. Mench2* 1 Fuel Cell Dynamics and Diagnostics Laboratory, and Department of Mechanical and Nuclear Engineering, The Pennsylvania State University, PA, USA, 2Electrochemical Energy Storage and Conversion Laboratory, and Department of Mechanical Aerospace and Biomedical Engineering, The University of Tennessee, TN, USA

1. INTRODUCTION The Department of Energy (DOE) 2015 technical target requires that fuel cell vehicles should deliver 50% of rated power in 30 seconds from a cold start at 20 C, with less than 62.5 J/We parasitic energy input for start-up and shut-down [1]. With those targets, they should also be able to start from 40 C after being soaked at this temperature for 8 hours. DOE targets for an 80 kWe (net) fuel cell system for automotive applications are summarized in Table 6.1. Dozens of studies have been performed to examine various aspects of fuel cell material compatibility and performance degradation in freezing environments. The damage resulting from a frozen soak, from freeze/thaw (F/T) cycling, and from frozen start have been examined. The purpose of this summary is to explore the damage resulting from freeze/thaw conditions. A detailed summary of damage observed due to some aspects of a frozen environment from various published studies is shown in Table 6.2. Generically, damage resulting from a frozen environmental condition is due to water generated at the cathode during sub-zero operation, or the existence of liquid water that resides in the membrane, porous media after shut-down. Liquid water that freezes in the channels and internal manifolds can hinder cold start and reactant flow, resulting in exacerbated degradation due to cell voltage reversal and carbon corrosion or local fuel starvation. An important result from accumulated studies is that, for conventional fuel cell materials and designs, no significant damage is observed from simply cycling the fuel cell material to subzero conditions without start-up operation or liquid water before freeze. This indicates that freeze-related damage can be eliminated through proper Polymer Electrolyte Fuel Cell Degradation. DOI: 10.1016/B978-0-12-386936-4.10006-5 Copyright Ó 2012 Elsevier Inc. All rights reserved.

293

294

Polymer Electrolyte Fuel Cell Degradation

TABLE 6.1 DOE Technical Targets for Automotive Applications: 80-kWe (net) Integrated Transportation Fuel Cell Power Systems Operating on Direct Hydrogena. As Reported in [1] Units

2003 Status

2005 Status

2010

2015

Energy efficiency at 25% of rated power

%

59

59

60

60

Energy efficiency at rated power

%

50

50

50

50

Power density

W/L

440

500

650

650

Specific power

W / kg

420

470c

650

650

e

Characteristic b

d

Cost

$ / kWe

200

110

45

30

Transient response (time from 10% to 90% of rated power)

seconds

3

1.5

1

1

Cold start-up time to 50% of rated power at 20 C ambient temperature at þ20 C ambient temperature

seconds seconds

120 60

20 <10

30 5

30 5

Start-up and shut-down energyf from 20 C ambient temperature from þ20 C ambient temperature

MJ MJ

N/A N/A

7.5 N/A

5 1

5 1

Durability with cycling

hours

N/A

~1,000g 5,000h

5,000h

Unassisted start from low temperaturesi



N/A

20

40

C

40

a

Targets exclude hydrogen storage, power electronics and electric drive. Ratio of DC output energy to the lower heating value of the input fuel (hydrogen). Peak efficiency occurs at about 25% rated power. c Based on corresponding data from the DOE report to account for ancillaries. d Based on 2002 dollars and cost projected to high-volume production (500,000 systems per year). e Status is from 2005 TIAX study and will be periodically updated. f Includes electrical energy and the hydrogen used during the start-up and shut-down procedures. g Durability with cycling is being evaluated through the Technology Validation activity. Steady-state stack durability is 20,000 hours. h Based on test protocol issued by DOE in 2007. i 8-hour soak at stated temperature must not impact subsequent achievement of targets. b

design, materials, and operating protocol. Based on the summary of available studies shown in Table 6.2, not all environments, materials, or designs result in damage. In fact, there has been a seemingly high discrepancy between the results of different studies, which suggests there is still much to learn about the

Chapter | 6

TABLE 6.2 Summary of Observed PEFC Damage Due to Frozen Environments from Various Sources Test conditions

Wilson et al. 1994 [62]

In-situ F/T

Nafion 117 20 wt% Pt/C Decal processa (0.16 mg/ cm2)

ELAT 10/80 3 hydrophobic carbon cloth

McDonald et al. 2004 [17]

Ex-situ F/T

Nafion 112 0.4 mg Pt/C /cm2

N/A

None

In-situ F/T

Nafion 112 0.4 mg Pt/C /cm2

N/A

Carbon paper

Ex-situ F/T Nafion 112 N/A (immersion)

N/A

None

DSM

N/A

N/A

Nafion 112 N/A DSM

In-situ F/T

N/A

MEA

DM

Results

No purge (wet)

No performance loss

40/80 385

Dry state (l<3)

No significant physical damage change in the molecular level

385

Dry state (l<3)

No significant physical damage change in the molecular level

40/50 10

Immersed in water

Severe CL loss Severe deformation of MEA

None

10

Immersed in water

No observable loss

N/A

N/A

40

N/A

No performance loss No ECSA loss

N/A

N/A

40

N/A

No performance loss No ECSA loss

295

(Continued )

Freeze Damage to Polymer Electrolyte Fuel Cells

Test Mode PEM

Liu 2006 [2]

CL

T range Number Purge/no ( C) of cycles purge

Reference

296

TABLE 6.2 Summary of Observed PEFC Damage Due to Frozen Environments from Various Sourcesdcont’d Test conditions

Test Mode PEM

CL

MEA

DM

T range Number Purge/no ( C) of cycles purge

Results

Patterson et al. 2006 [63,64]

In-situ F/T

N/A

N/A

N/A

N/A

40/25 63

N/A

No performance loss

Cold start-up

N/A

N/A

N/A

N/A

15

N/A

End cell loss

Mukundan et al. 2006 [20,68]

In-situ F/T

Nafion 1135 20 wt% Pt/C Decal (0.2 mg/cm2) processa

Wet proofed 40/80 100 carbon cloth

No purge (wet)

No performance loss

SGL 30DC

No purge (wet)

Mechanical failure of DM

Cho et al. 2003, 2004 [21,65]

In-situ F/T

N/A

45

Nafion 1135 20 wt% Pt/C (0.2 mg/cm2)

Wet proofed 80/80 10 carbon cloth

No purge (wet)

Performance loss HFR increase Interfacial delamination DM failure

Nafion 115 20 wt% Pt/C GDEc (0.4 mg/cm2)

Wet proofed 10/80 4 carbon paper

No purge (wet)

Performance loss, ohmic and charge transfer resistance increase ECSA loss

4

Dry purge (l<2)

No performance loss No ECSA loss

Polymer Electrolyte Fuel Cell Degradation

Reference

N/A

Meyers 2005 [15]

In-situ F/T

Commercial MEAs (reinforced membrane)

N/A

20/?

Oszcipok et al. 2005, 2006 [66,67]

Cold start-up

Catalyst coated membrane

N/A

10

Catalyst coated membrane (0.4 mg Pt/cm2)

Carbon cloth 10 GDEd

Carbon paper Exposed N/A to freezing

Carbon paper /cloth

Yan et al. 2006 [14]

Cold start-up

Guo and Qi 2006 [18]

Ex-situ F/T Commercial MEA with 30 mm membrane None and 1.0 mg Pt/cm2

In-situ F/T

Nafion 20 wt% 112,115,117 Pt/C

N/A

Carbon paper

15

N/A

DM Fracture Membrane failure Severe CL delamination

20

N/A

Membrane cracks CL delamination

10

Dry purge

Performance loss ECSA loss Hydrophobicity loss (MOL,DM)

7

Partial purge

Significant performance loss

N/A

30/20 6

20

Interfacial delamination Membrane hole Dry purge (l<4)

Negligible damage

No purge (wet)

Severe damage Severe CL cracks

No purge (wet)

Severe CL cracks ECSA loss Negligible performance loss Easy flooding

Dry purge (l<4)

No physical damage No performance loss

297

(Continued )

Freeze Damage to Polymer Electrolyte Fuel Cells

Field test N/A (stationary)

Chapter | 6

Gaylord 2005 [16]

298

TABLE 6.2 Summary of Observed PEFC Damage Due to Frozen Environments from Various Sourcesdcont’d Test conditions

Reference

Test Mode PEM

Hou et al. 2006 [19]

In-situ F/T

Alink et al. 2008 [31]

In-situ F/T

CL

MEA

Nafion 212 20 wt% Pt/C GDE (0.8 mg Pt/ cm2) 0.4 mg Pt/C /cm2

DM

T range Number Purge/no ( C) of cycles purge

Carbon paper 20/60 20

Toray TGP-H- 40/60 120 060

20/0.5 10

Ex-situ F/T Cold start-up

0.4 mg Pt/C /cm2 0.3 mg Pt/C /cm2

Toray TGP-H- 40 060 9

No performance loss No ECSA loss No physical damage Dry purge

Increase in porosity of MEA Decrease in electrode surface area more important at the cathode Micro-cavities on electrodes

No purge

Increase in porosity of MEA Serious detachment of electrode material Micro-cavities on electrodes No damage to MEA No damage to DM

Polymer Electrolyte Fuel Cell Degradation

62

0.3 mg Pt/C /cm2

Results

Ex-situ F/T

Non-cracked Reinforced 0.4 mg Pt/C membrane /cm2 18 mm

Without DM/ 40/70 30 MPL

CL separation under channels

Reinforced membrane 18 mm

Severe CL separation under channels

Noncracked 0.4 mg Pt/C /cm2

Nonreinforced 18 mm

Nearly complete delamination of CL

Noncracked 0.4 mg Pt/C /cm2

Reinforced 35 mm

Severe MEA damage, nearly complete delamination of CL under channels

Noncracked 0.4 mg Pt/C /cm2

Reinforced membrane 18 mm

Cracked 0.4 mg Pt/C /cm2

Reinforced membrane 18 mm

Non-cracked Non0.4 mg Pt/C reinforced 18 mm /cm2

With DM/ MPL

40/70 30

MEA largely intact. No sign of F/T damage

Cracking with no delamination

Freeze Damage to Polymer Electrolyte Fuel Cells

Cracked 0.4 mg Pt/C /cm2

Chapter | 6

Kim et al. 2007 [3]

Damage under channels

299

(Continued )

300

TABLE 6.2 Summary of Observed PEFC Damage Due to Frozen Environments from Various Sourcesdcont’d Test conditions

Reference

Ex-situ F/T

CL

MEA

Noncracked 0.4 mg Pt/C /cm2

Reinforced 35 mm

DM

T range Number Purge/no ( C) of cycles purge

Results Frost heave damage and CL separation

Carbel-CL SGL 10BB SGL 25BC

40/70 30e100

Water Interfacial delamination submerged DM/CL condition Deformation of stiff diffusion media

SGL 10BA Decal printing (TBAþ form catalyst) and then hot pressing at 200  C. 20% PTFE treatment with MPL. Abbreviation: F/T: Freeze/thaw thermal cycling DSM: Dimensionally stable membrane GDE : Gas diffusion electrode, PEM: Polymer electrolyte membrane. c Catalyst ink sprayed on DM and then hot pressing at 140 C. d Sprayed on DM and then hot pressed. a

b

Polymer Electrolyte Fuel Cell Degradation

Kim et al. 2008 [4]

Test Mode PEM

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

301

genesis and causes of freeze related damage. There are some commonalities between the studies. When freeze related damage was observed, it was generally observed at the following locations: l l

l

The membrane in the form of pinholes on the surface. The catalyst layer (CL), in the form of local cracks as well as interfacial delamination between the CLjmembrane and/or CLjDM interface, and loss of electrochemical surface area (ECSA). The diffusion media, via cracking of the microporous layer (MPL) and interfacial delamination with the CL, or loss of hydrophobicity.

The effect of different component characteristics on sustaining damage was studied [2,3,4]. One important result concludes that properly drying a cell before sub-zero cool down and freeze prevents observable physical damage. However, over-drying or non-uniform drying of the membrane during shut-down has been shown to cause an uneven stress distribution in the membrane and result in accelerated membrane degradation [5]. Although freeze-related damage can be eliminated by completely purging the cell, this mitigation method is generally too time consuming, parasitic, and potentially damaging to the membrane to be of use. Much more study of purge, evaporative removal, and knowledge of the locations and sources of freeze-related water damage is needed. From Table 6.2, it can be seen that damage is not uniformly observed and varies as a function of: 1. Material sets – Microporous layer (MPL)jCL combinations appear to impact results. This is a function of drainage at shut-down, interfacial contact, and distribution of compression pressure. Stiff or bonded DM appear to best mitigate damage because they can reduce interfacial liquid accumulation which has been shown to cause freeze delamination in some cases [3,4]. 2. Cell design – The channel/land design also appears to impact damage. In particular, for a traditional channel/land configuration, low under-channel compression exacerbates damage. The wider the channel, the worse the damage appears to be. Figure 6.1 shows the result for freeze/thaw cycling to 30 C from a wet state with 2 mm wide lands. This testing qualitatively confirmed the results of computational simulation [6–8] which predicted that the major damage locations from a macroscopic perspective appear under the channel along various material interfaces in the cell. 3. Shut-down protocol – Obviously, the source of damage is water. Upon cooldown, condensation, diffusive flow, temperature gradient driven flow, and capillary transport take place. To avoid damage or hindered cold-start, residual water inside the CL should be removed. This can be accomplished with a variety of methods discussed in Section 4 of this review. 4. Location in stack – Anode end cells in particular have been shown by several groups to suffer aggravated damage compared to center and cathode end stack plates. This is apparently a result of greater heat transfer from end plates and concomitant phase-changed induced (PCI) motion [9].

302

Polymer Electrolyte Fuel Cell Degradation

Channel

Land

FIGURE 6.1 SEM image of cross-section of membrane electrode assembly after 100 freeze/thaw cycles to 30 C. Delamination damage is seen under the channel but not land locations due to the overburden pressure from the land [4].

Although many studies have been conducted to investigate freeze damage on different components of a polymer electrolyte fuel cell, there are significant variation in the results. Much, but likely not all, of this variation can be ascribed to the non-standardized testing procedures, cell designs, and materials used. There is still a clear need to resolve these discrepancies with fundamental understanding of the physicochemical mechanisms involved, so that optimized designs, materials, and protocols can be developed. The motivation of this review is to better understand and codify the existing data so that conflicting conclusions from the various studies can be better explained. Additionally, methods and potential concepts to mitigate freeze-induced damage are discussed. The operation of start-up from a frozen state and possible damage resulting exclusively from a frozen start is another topic that is not within the scope of this review.

2. COMPUTATIONAL MODEL EFFORTS Several publications based on computational models for predicting the key parameters and conditions for freeze damage in PEFCs have been developed. These are based on porous media flow theories, and frost heave formation in thin cracks [6–8]. Models to predict the formation of ice, without the onset of damage, have also been developed, but are not within the subject of this review and are not summarized here. Damage resulting from ice formation can be separated into two different phenomenological categories: 1. damage due to ice formation and expansion from a liquid to solid state; and 2. damage due to an ice lens, which develops sufficient phase pressure to physically separate interfacial surfaces. Damage due to the approximately 8% volume expansion of the ice phase is imagined by many to be the only possible mechanism for damage, but

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

303

experimental and computational evidence now suggests that ice lens formation is responsible for much of the observed freeze damage. In fact, the porous media in fuel cells (CL and DM) are typically far from a full-saturation state, and another 8% expansion as the ice forms during a slow cooling at shutdown is not likely to cause severe morphological damage. In contrast, an ice lens can cause plastic or elastic delamination and deformation, and can form as a result of even slight interfacial water accumulation, or from water expulsion from the membrane under decreasing temperatures [6–8]. The presence of interfacial accumulation of liquid has been suggested or observed by several independent studies [10–12]. After ice nucleation, whether or not an ice lens continues to grow depends on the ice phase pressure, the overburden pressure, the hydraulic availability of liquid water flowing to the ice lens location, and the heat transfer rate from the porous media. The critical ice pressure for ice lens formation is referred to as the local overburden pressure Povbd. Overburden pressure is a function of the assembling pressure Passm or the transmitted channel pressure Pch, depending on whether the location is under the land or the channel respectively, the material tensile strength sts and the shear stress ssh [6]. The possible cases are listed in Table 6.3. The impact of DM stiffness is discussed in the following section of this review. The large overburden pressure restrains macroscopic ice lens formation. However, local delamination at the catalyst level could still occur due to high local ice-phase pressure. Figure 6.2 is a schematic representing potential locations of freeze/thaw damage according to a computational model, which included the impact of water motion in the ionomer and porous media during shut-down to a frozen state [6–8]. The results are in qualitative agreement with the observed F/T damage on SEM for materials which underwent ex-situ F/T cycling. Ice growth leading to damage most likely occurs under the channel and at the interface between CLjDM and CLjMembrane. The CLjMembrane ice growth is highly dependent on the freezing temperature depression properties of Nafion membrane. Non-freezing water flowing out of the membrane would immediately freeze upon contact with the catalyst layer. The maximum ice lens growth at this location would therefore depend on the initial water content of

TABLE 6.3 Possible Locations of Ice Lens Growth with their Respective Overburden Pressure Based on Data from [6] Position

Under BP

Under CH

Within DM, CL or Nafion

Povbd ¼ Passm þ sts þ pch

Povbd ¼ pch þ sts þ ssh

At interface between bipolar plate(BP)/DM, DM/CL or CL/Nafion

Povbd ¼ Passm þ pch

Povbd ¼ pch þ ssh

304

Polymer Electrolyte Fuel Cell Degradation

FIGURE 6.2 Schematic showing the potential locations of freeze/thaw damage [6].

the membrane, and the membrane type, as represented in Fig. 6.3. Reduction of the liquid water in contact with the ionomer at shut-down to a frozen state is the key to avoid damage, as the liquid contact is responsible for higher free-water content in the membrane, which is responsible for local degradation – as discussed. It should be noted that extensive ex-situ testing revealed that no damage was observed when F/T cycling in a purely gas phase (but vapor-saturated) environment [13].

3. MODES OF DEGRADATION In this section, the various modes of observed freeze-related physicochemical damage are discussed. Where possible, the phenomena responsible for the damage are described. The section is divided into subsections based on

FIGURE 6.3 Maximum thickness of ice lens that could be formed by water expelled from Nafion during freezing, as a function of the initial water content and membrane type [6].

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

305

components for convenience, although it is likely that the observed damage is interrelated.

3.1. Membrane Water freezing in the PEFC can damage the electrolyte membrane in different ways. Physical damage is observable with a scanning electron microscope (SEM). Increased roughness, cracks and pinholes were observed in a membrane after in-situ operation at sub-zero temperatures ranging from 5 C to 15 C [14]. The same was observed of cells in a fuel cell stack that were freeze/thawed 20 times at 20 C [15], and idle stacks during winter or long duration installation under freezing temperatures [16]. Figure 6.4 shows SEM images of damage to a Nafion membrane after being stored and operated at sub-zero ambient temperatures as low as 15 C [14]. The MEA was assembled by spraying the catalyst on wet-proofed carbon paper and then hot pressing the electrodes on the Nafion. After operation the electrodes were separated from the membrane and the polymer electrolyte was examined with an SEM. Increased membrane roughness is observed (Fig. 6.4(c)) after sub-zero operation compared to a membrane operated at room temperature (Fig. 6.4(b)). At higher magnification, (Figs. 6.4(d) and (e)) microcavities and pinholes at the cathode outlet region of the membrane are clearly visible after sub-zero operation. This type of damage can lead to performance loss through increased hydrogen crossover and loss of catalyst activity. In the electrolyte, water content (l) is defined as the number of moles of water per mole of sulfonic acid group in the electrolyte. Membrane hydration is elementary to ionic conductivity, but over-hydration and subsequent freeze can cause damage. A dried MEA with l < 4 (water weight percent less than 6%), did not experience freeze-damage [17–20]. However, this state has difficulty in generating current from a frozen state because of low ionic conductivity [21,22]. Under sub-freezing conditions, Mukundan et al. [20] determined that 7
306

Polymer Electrolyte Fuel Cell Degradation

(a)

(b)

(c)

(d)

(e)

FIGURE 6.4 Effect of sub-zero temperature on membrane (a) Virgin Nafion membrane, (b) membrane after operation at room temperature, (c) membrane after operation at 15 C, (d) membrane from cathode outlet regions after operation at 15 C and (e) membrane from cathode outlet regions after operation at 15 C. Images from [14].

grains are, the greater the freezing point depression compared to free standing water. DSC data from the literature [24,25,26] characterizing water composition in Nafion has been extrapolated by He and Mench [6] and is shown in Fig. 6.5. It was shown that the weakly bonded water in Nafion pores sized 2 nm corresponds to a freeze point depression of 24.5 K. It was hypothesized by He et al. that this weakly bound water comes out of the membrane and results in freezing damage at the interface between the membrane and the catalyst layer. An

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

307

FIGURE 6.5 Unfrozen water versus temperature curves derived from Nafion DSC data of ref 24,25,26. [6].

experimental study by Pineri and co-workers appears to confirm this [27]. In the Pineri work, X-ray diffraction results indicate that some of the water desorbs out of the membrane below 0 C. Damage could also be a result of the membrane swelling and contraction with temperature. The Pineri result suggests that the damage is more likely a result of water outflow than membrane swelling. The end result is the same for either case, and indicates that a major source of potential damage is excess membrane water resulting from liquid water in contact with the electrolyte at shut-down to a frozen state. A study by Liu measured the strain before failure of Nafion 112 membrane after dry and wet F/T cycles. The results are summarized in Table 6.4. The membrane after dry F/T cycles from 40 C to 80 C ruptures under significantly less strain than a membrane not subjected to F/T, indicating some internal change in structure. Wet F/T cycling did not show any additional TABLE 6.4 Percent Elongation at Break of Nafion 112 Membranes Before/ After Dry/Wet F/T Thermal Cycling Compiled from [28] Material

Percent Elongation at break Before Cycling

After 385 F/T Cycles (Dry)

Membrane (Machine direction)

1290

40

Membrane (Cross direction)

320

25

Before Cycling

After 200 F/T Cycles (Wet)

Membrane (Machine direction)

>300

>300

Membrane (Cross direction)

>300

>300

308

Polymer Electrolyte Fuel Cell Degradation

potential, as the strain after 200 wet F/T cycles was the same as the strain of a new membrane [28]. The authors suggest that water in the membrane relieves structural change that can occur during freezing, because it makes chain movements more facile. It therefore prevents the membrane from becoming brittle. This effect may also contribute to the observed exacerbated damage from uneven dry-out during purge of large stack plates. Areas of high local dryout can suffer from reduced plasticity in the membrane on subsequent purges.

3.2. Catalyst Layer Damage Maintaining catalyst layer integrity throughout operation is of critical importance to a fuel cell performance, since other non-freeze related degradation modes commonly cause significant irreversible damage to the catalyst, membrane, and support structure [29]. The catalyst layer (or electrode) is a porous media covering both faces of the membrane, with a typical thickness range of 5–30 mm and porosity of 0.4–0.6. A surface morphology characterization of catalyst layer was performed by Hizir et al. [30]. Local catalyst cracks with large relative dimensions O(mm) compared to pores are often observed after F/T and sub-zero operations of fuel cells. Because the catalyst layer is between the membrane and the gas diffusion media, interfacial CLjmembrane and CLjDM delamination is often observed due to ice lens formation. In addition, because the catalyst layer is a reaction site, any damage to will most likely lead to a loss in electrochemically active area. However, there exist conflicting results, as some researchers observed damage at the electrodes in the form of lost ECSA, physical cracking or pulverization of the electrode, while other studies did not show any damage. The work of Kim et al. [3,4] investigated those conflicting conclusions by studying the effect of fuel cell component structures, DM stiffness/thickness and membrane rigidity on the impact of freeze thaw (F/T) damage on the electrodes. Kim et al. determined that a stiff DM with a thin, reinforced membrane was the best configuration to mitigate damage from a freeze/thaw environment. DM thickness was not found to play a significant role in freeze/thaw damage.

3.2.1. Electrode Cracking Extensive cracking of the electrode structure has been observed on MEAs subject to freeze-thaw cycles in both ex-situ and in-situ conditions. Figure 6.6 is an SEM image from the work of Guo and Qi and depicts the change in surface morphology of a commercial MEA frozen after being subjected to ambient temperature and RH (Fig. 6(a)) versus a similar MEA that was frozen after being hydrated in water at 80 C for 10 minutes (Fig. 6(b)). Each were cycled six times between 20 and 30 C and soaked at 30 C for 6 hours during every cycle [18]. In Fig. 6.6(a) the electrode is smooth and almost no damage can be seen as a result of the six freeze thaw cycles. However, severe damage is apparent in Fig. 6.6(b) with cracks and obvious separation of the catalyst

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

(a)

309

(b)

FIGURE 6.6 SEM of the cathode side of freestanding MEAs after six freeze-thaw cycles between 20 and 30 C: (a) MEAwas only exposed to ambient temperature and relative humidity before going through the freeze-thaw cycles; 100x magnification; (b) MEA that was fully hydrated in water at 80 C for 10 min before going through the freeze/thaw cycles; 50x magnification. Images from [18].

surface. In the upper left corner, a total detachment of the catalyst material from the membrane can also be seen. Alink et al. also made a similar observation exsitu with freeze/thaw cycling down to 40 C of a wet gas diffusion layer and MEA assembly without compressive forces [31]. In fact, without assembly compression the mechanical bond is very weak and repeated testing from various groups shows that detachment readily occurs. Alink et al. also used a commercial type MEA for in-situ freeze/thaw cycling with compressive forces. One assembly was exposed to fully humidified reactants at the cathode and anode before being cycled. SEM showed catalyst damage and fracture, although catalyst layer segregation was not as noticeable as in the ex-situ experiment. This could be due either to the fact that an MEA in an assembled fuel cell is subjected to pressure from the backing plates, or the fact that the membrane water uptake is less when subjected to vapor instead of liquid [31] – or indeed both. This is in agreement with other research that explains how water drains and freezes in the catalyst layer below 0 C when membrane water content has reached a maximum [31]. This is also due to the fact that some water inside the membrane does not freeze, as explained in Section 2. Additionally, upon freezing in the electrode pores, water expands in volume and may generate micro-cracks on the surface of the catalyst [18,20,31].

3.2.2. Interfacial CLjMembrane & CLjDM Delamination Delamination of the catalyst layer from both the membrane and the DM sides was shown to occur in several publications after the cell is subjected to sub-zero operation or brought to a frozen state without effective purging [3,4,14–16,20]. The catalyst layer and the DM are both porous media and permeable to gas. With pore sizes ranging from several nanometers to hundreds of micrometers,

310

Polymer Electrolyte Fuel Cell Degradation

(a)

(b)

(c)

(d)

FIGURE 6.7 Effect of sub-zero temperature on MEA with cloth DM. (a) Virgin MEA, (b) MEA after operation at room temperature, (c) MEA after operation at 15 C. Images from [14].

water confined in those pores experiences a freezing point depression of only 2 to 4 C, which is not enough to prevent freeze damage [32]. The cathode side is more prone to separation as water is generated in the cathode catalyst layer by the oxygen reduction reaction. Figure 6.7 shows the evolution of a virgin MEA (Fig. 6.7(a)) upon operation at room temperature (Fig. 6.7(b)) where no delamination is observed from liquid water. Delamination on both the membrane and DM side became apparent after operation at 10 C (Fig. 6.7(c)) and 15 C (Fig. 6.7(d)). The delamination normally occurs under a channel location, where overburden pressure is low. For this reason, open or mesh flow field designs have an intrinsic advantage over conventional channel land design for limiting freeze/thaw damage. To better understand CL delamination and how fuel cell components can help to promote or mitigate freeze damage, Kim et al. investigated the effect of DM stiffness, DM thickness and membrane rigidity on freeze/thaw damage in an ex-situ environment. The results are summarized below. Effect of DM Stiffness Frost heave formation and volume expansion of frozen water can induce shear force on the catalyst layer leading to interfacial delamination. Frost heaving is

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

311

a phenomenon more complex than volume expansion of frozen water. Primary and secondary heave can occur whether or not an ice fringe exists, depending on the thermal and mass transport conditions. As shown in Fig. 6.8, the frozen fringe is the transition two-phase zone between 0% and 100% ice at the freezing front. Primary heave refers to frost heave with no frozen fringe, where no ice will penetrate into the unfrozen area. During secondary heave, the ice lens grows and penetrates into the frozen fringe [16]. Two test cells, one with flexible cloth DM and another with stiff carbon paper type DM, were F/T cycled 30 times from 40 to 70 C [4]. They are both shown in Fig. 6.9. Excessive surface damage and CLjDM delamination were observed on the catalyst layer of the cell assembled with cloth DM (Fig. 6.9(a)), while no cracks were observed on the cell using stiff carbon paper DM (Fig. 6.9(b)). A stiffer diffusion media more uniformly translates the compressive forces from under the land to under the channels and therefore provides more deformation resistance when subject to ice growth pressure. The stronger compression to the CL surface can also reduce interfacial water accumulation at shutdown. This also means that the channel width and channel/land ratio are important parameters in uniformly spreading the compressive forces; relatively wide channels will promote DM deformation. Figure 6.10 shows a calculated compression distribution from a common felt DM (SGL 10BB) onto the CL. To obtain the non-homogeneous compression pressure data required for the simulation, the DM thickness versus compression pressure data given in [33] is used to evaluate the non-homogeneous strain in the DM layer. Finally, using the DM strain data, the compression information under one land-channel configuration is extracted from the stress-strain data of the DM given in [34]. The DM thickness measurement was performed ex-situ for one set of land and channel (each of length 1 mm), and Fig. 6.10 shows the variation of the non-homogeneous compression pressure from mid-land to mid-channel location with a span of 1 mm. As can be seen, even for a stiff DM, the compression pressure on the catalyst layer drops off very sharply in the channel

(a)

(b)

FIGURE 6.8 Comparison of (a) primary and (b) secondary frost heave. Images from [6].

312

Polymer Electrolyte Fuel Cell Degradation

(a)

(b)

FIGURE 6.9 Surface images of MEAs cycled 30 times between 40 and 70 C with negligible cracks in the virgin catalyst layer and 18 mm reinforced membrane. Images shown correspond to locations under channel. (a) CARBEL-CL (cloth type) DM and (b) SGL 10BB (non-woven felt type) DM. Images from [4].

region. This is the reason delamination damage is most likely in this location. Based on modeling from S.He et al, the calculated ice-phase pressure rarely gets over 2 MPa, which is at the high range of the normal compression experienced under a land in a typical fuel cell. Effect of DM Thickness Kim et al. also compared the effect of carbon paper DM thickness on a 35 mm reinforced membrane known to be sensitive to damage [4] and an 18 mm

Compression pressure (MPa)

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0

0.2

0.4

0.6

0.8

Mid-land to mid-channel span (mm) FIGURE 6.10 Calculated compression distribution from a common DM onto the CL.

1.0

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

313

reinforced membrane known to be less sensitive, as seen in Fig. 6.11. The non-woven felt type is stiffer due to its more three dimensional lattices. Both cells with 18 mm membrane showed no physical damage with either thin carbon paper DM (235 mm) of non-woven paper type (Fig. 6.11(a)) or thicker carbon (415 mm) non-woven felt type DM (Fig. 6.11(b)) respectively. This is a result of stiff DM applying some compressive force under the channels. Interestingly, the stiff DM did not prevent damage on the thicker membrane (Fig. 6.11(c)) and the thicker stiff DM showed as much damage (Fig. 6.11(d)). It was concluded that a thickness from 235 mm to 415 mm of non-woven type was not significant to mitigate the observed physical damage to the electrode surface [4]. Effect of Membrane Rigidity and Thickness A test cell with 18 mm non-reinforced membrane and carbon paper DM was F/T cycled 30 times between 40 C and 70 C [3]. Fig. 6.12 shows interfacial delamination under the channel. Although the DM was stiff carbon

(a)

(b)

(c)

(d)

FIGURE 6.11 Cross-sectional images of MEAs with negligible virgin cracked catalyst layers, F/T cycled 30 times: (a) 18 mm reinforced membrane with SGL 25BC DM (thickness 235 mm); (b) 18 mm reinforced membrane with SGL 10BB DM (thickness 415 mm); (c) 35 mm reinforced membrane with SGL 25BC DM; (d) 35 mm reinforced membrane with SGL 10BB DM. Images from [4].

314

Polymer Electrolyte Fuel Cell Degradation

FIGURE 6.12 SEM image of F/T cycled non-cracked CL with 18 mm non-reinforced membrane under the channel location [3].

paper, the non-reinforced membrane promoted delamination. Membrane reinforcement is used to make a membrane mechanically stronger and more durable without significantly changing its conductive capabilities. The most common methods include adding a strong polymer such as expandable porous polytetrafluoroethylene or other fibers, resulting in a membrane composite [35]. When using a thicker 35 mm reinforced membrane with carbon paper DM, as shown in Fig. 6.13, frost heave damage is visible under the channels. Although the thicker membrane is reinforced, it is a bigger

FIGURE 6.13 SEM image of F/T cycled non-cracked CL with 35 mm reinforced membrane under the channel location [3].

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

315

reservoir for water and by itself a source of water for damage in the CL [3]. Therefore, the best material combination to mitigate freeze-damage was found to be a non-cracked virgin catalyst layer on a reinforced, thin membrane, assembled with stiff diffusion media. Although this freezetolerable design reduced freeze damage under worst case scenarios of direct liquid contact with the ionomer at freeze, irreversible damage was still present, highlighting the importance of liquid removal from the catalyst layer before shut-down to a frozen state.

3.3. Loss of Electrochemical Surface Area Besides physically observable damage, performance is directly relevant to the electrochemical surface area (ECSA) at the electrodes, which can be measured in-situ with cyclic voltammetry [18,20,32,36]. Even without major observable morphological damage, ECSA loss has been observed in F/T testing. After 20 freeze/thaw cycles (20 to 30 C at fully humidified state) Guo and Qi observed ECSA loss at both electrodes. ECSA decreased by 23% at the cathode and by 15% at the anode, as seen in Fig. 6.14. This difference could be due to storage of generated water in the cathode CL due to previous operations. However, the short term performance of the cell did not show much change [18]. Hou et al. investigated freeze degradation using 20 freeze/thaw cycles between 20 C and 60 C. The cell was operated at 60 C, purged by gases at 25 C with 58% RH after each operation, and then frozen to 20 C. Cyclic voltammetry (performed only at the cathode) showed that values of ECSA fluctuated between 45.4 and 51.2 m2/ gcat and did not decrease progressively after each cycle [19]. Interestingly, this fluctuation did not alter the performance curves after each freeze/thaw cycle. Although the ECSA measurement fluctuation could be from the experimental device, cyclic voltammetry is a transient test and it is possible to have detected structure alteration or liquid water transients blocking the triple-phase boundary which would not affect a steady-state performance test. Mukundan et al. [20] also performed ECSA measurements to compare the durability of their Los Alamos National Lab (LANL)-made MEAs to MEAs from W.L. Gore. The W.L. Gore MEAs showed >50% loss in the catalyst surface area after five cold starts at 10 C, while LANL-prepared MEA showed negligible loss. In this study, catalyst layer morphology is obviously important for durability at freezing conditions; as previously discussed, water in smaller pores may not freeze at conditions in which water in larger pores does. This loss in ECSA after cold starts was not observed at the anode, clearly because water is generated at the cathode side. Ge and Wang [32] and Srouji [36] made the same observation regarding lack of damage at the anode from cold starts. Ge and Wang have also recorded 1 to 3% of Pt area loss at the cathode per cold start performed at 10 C, although each cold start was interrupted by a thaw and operation at 70 C, making the cell go to a freeze down process before each cold start [32]. Srouji recorded 4.4% of Pt area loss at the cathode after 25 consecutive cold starts at 10 C. The protocol

316

Polymer Electrolyte Fuel Cell Degradation

FIGURE 6.14 Cyclic voltammograms of (a) cathode and (b) anode of an MEA after 0–20 freeze/thaw cycles at fully humidified state. Images from [18].

developed for rapid consecutive cold starts with known initial membrane water content is described in detail by Chacko et al. [37], and is capable of isolating sub-zero operation damage from the water generated at the cathode from residual water damage resulting from the freeze down process itself. A challenge in cyclic voltammetry studies is to correlate ECSA loss with performance loss during steady-state operation.

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

317

3.4. DM Fracture and Loss of Hydrophobicity Some failure of the gas diffusion media has been observed as an apparent consequence of F/T cycles. Mukundan et al. [20] observed DM failure after 10 F/T cycles down to 80 C. However, Yan et al. [14] witnessed no damage to DM after exposure to normal conditions but noticed an increase in porosity and darkness in color after sub-zero temperature exposure. They attributed this phenomenon to ice forming in the DM. Results from neutron imaging studies have shown that the saturation of the DM materials rarely can exceed 30%, so that the additional 8% volume expansion from freezing should be tolerable in the overall structure. However, some damage can occur if the water is locally confined by an enclosed pore structure or surrounding ice. Although the DM provides a stiff support for the MEA and a hydrophobic barrier, it does store a considerable amount of water in the CL after shut down [31]. Evidence of reduced hydrophobicity from exposure to freezing conditions with high liquid saturation has also been observed. The damage to the DM in freeze conditions needs to be investigated more for a more complete understanding.

4. METHODS OF FREEZE DAMAGE MITIGATION There are various concepts for preventing the fuel cell system damage caused by freezing, as well as a damage-free rapid start-up in sub-freezing conditions through good energy/power management. Pesaran et al. [38] categorized a review of solutions into two strategies: ‘Keep Warm’ where the system uses energy during vehicle parking and ‘Thaw and Heat at Startup’ which consumes energy mostly at vehicle startup. A summary of the various approaches is shown in Fig. 6.15. Intellectual properties have been developed for most if not all of them, and a compilation of 160 patents for freeze damage mitigation are listed and summarized in reference [38]. In that same milestone report, it is concluded that the correct use of insulation around the stack components can delay stack freeze by several days after it is shut down. Residual water reduction and evaporation during shut-down before the fuel cell is frozen can be achieved by several methods, including: 1. 2. 3. 4. 5.

convective purge; vacuum purge; capillary drainage; thermally driven drainage; and combinations of the above.

No more than 62.5 J/We should be consumed during cold start-up, based on the DOE goals for parasitic losses. The ultimate goal is a non-parasitic shut-down with no damage. Clearly, the key to shut-down is proper removal of liquid which is in contact with the ionomer without producing overly dry areas of the membrane. Several

Insulation Box

318

• Insulation

Vacuum Insulation • Energy/Power Source

Catalytic Burner Low-Power Stack Operation

Keep-Warm During Dwelling (Moist Parking)

Electric Heater • Sensing & Control

Temperature Sensing Freezing Judgment Controller

Determining Strategies Between the Two

• Strategy for Using Energy Energy Use Minimizing

Vacuum Drain

Thawing and Heat Up at Start (Drain Parking)

• Re-Humidification

Humidifier

• Internal Heating

Combusting H2 in Channels Reaction Heat at Catalyst Wire Heating in MEA

• External Heating Hybrid: FC & High Power ES (Battery/Ucap) FIGURE 6.15

Coolant Heating Hot Air Blowing

• Component

Thawing Tank Burner, Compressor

Method and technology chart for fuel cell start from subfreezing environment [38].

Polymer Electrolyte Fuel Cell Degradation

• Water Draining & Purging Discon. Reactants Humid.

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

319

studies agree that an MEA equilibrated to 80~95% relative humidity is better suited for rapid cold start, since this provides some storage for generated water during start-up and assures limited liquid-phase ionomer contact at freeze. However, this is difficult to achieve in practice without an exceedingly long purge, high parasitic losses, or distributed stresses which can lead to degradation and the fact that it is progressively more difficult to remove water with decreasing temperature, to maintain a dryer than saturated state [39]. A suggested optimal purge strategy is to keep purging until the water in the channels and diffusion media is removed, while water is still largely present in the membrane [5]. An MK 9 series 10 cell stack used in this study had an optimal purge duration of 88 seconds with dry air and H2 at 89 L/min and 25 L/min respectively; both at 70 C and 1.6 bar. However, this can be difficult to achieve in full size stack plates. Cho and Mench showed that for certain conditions, the water content in the cell is not correlated with high frequency resistance (HFR) during purge, and is not a good metric of water removal from the cell. Figure 6.16 shows this [40]. For this plot, data were taken using neutron imaging to record total liquid water content, and HFR, to record average membrane resistance. Different combinations of anode and cathode inlet relative humidity were used to purge a 250 cm2 full size fuel cell stack plate. Each test began from the same initial conditions. All comparative purges were operated at the same flow rates relative to each other. As can be seen from the figure, a full humidity (100/100% RH anode/cathode) purge results in water removal from the cell, indicating significant accumulation in the channels can

FIGURE 6.16 Cell water amount from neutron radiography with respect to membrane resistance at different operating conditions during purge [40].

320

Polymer Electrolyte Fuel Cell Degradation

be removed due to non-evaporative effects such as shear. Due to back diffusion and the initial water distribution, there is a sharp difference in the membrane dry-out compared to a dry anode or cathode purge. As discussed, membrane non-uniformities in water content have been shown to exacerbate damage and should be avoided. Recent work has shown a novel composite purge approach can most efficiently remove water content while preventing membrane dry-out [41,42], as shown in Figs 6.17(a) and (b). The characteristic water removal behavior during gas purge was analyzed using neutron radiography (NR) and HFR, as shown in Figs 6.17 (a) and (b). NR is used for quantifying the total

Liquid water amount (kg m-3 x 10-3)

(a)

50 High purge flow rate Medium purge flow rate Low purge flow rate Composite purge flow rate

47 44 41 38 35

0

3

6

9

12

15

12

15

Purge time (minute)

(b)

12 High purge flow rate Medium purge flow rate Low purge flow rate Composite purge flow rate

Cell resistance (m • )

10 8 6 4 2 0

0

3

6

9

Purge time (minute) FIGURE 6.17 Water removal behavior of fuel cell during purge: (a) variation of water amount in the cell and (b) variation of total cell resistance [41].

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

321

amount of water residing in all the components of the fuel cell, whereas HFR is utilized to indicate variation of water content in the membrane. Therefore, by comparing both data sets during purge, water removal behavior can be understood in detail. As shown in Figs 6.17 (a) and (b), a high flow rate purge was very fast and efficient for decreasing the residual water in the cell, but increased the cell resistance substantially, raising issues of possible degradation of the membrane and high energy consumption. For a relatively low flow rate purge, the cell resistance did not increase severely, but water removal from the cell was not efficient. However, in the case of a composite purge with mixed purge flow rates (high flow rate for 1 min., medium flow rate for 3 min., and low flow rate for 10 min.), the water removal rate from the cell was almost identical to the medium flow rate case, but with reduced membrane resistance increase (91%) and less energy consumption (24%). More details of this can be found in ref. [41]. A typical convective method of removing residual water during shut-down is purging with hot dry gas, which is effective in rapidly evaporating residual liquid water from inside the DM or CL and the channels. The convenience of this method depends on the reactant gas flow field patterns. It often leads to non-uniform water distribution, which can result in rapid degradation of the MEA. For example serpentine flow field patterns have more water content near the outlet and suffer from dry out at the gas inlets. This results in mechanical stress causing physical degradation. Serpentine flow fields also suffer from water accumulation around the 180 turns, as shown in Fig. 6.18. This accumulation would tend to damage the cell upon freeze or prevent proper start-up via channel blockage. However, parallel flow fields have less resistivity to fluid motion and hence mitigate non-uniformity [43]. This general effect has been observed consistently in both small and full size stack designs, leading to a modern design paradigm that seeks to straighten the flow field as much as FIGURE 6.18 Neutron radiograph showing a tendency for water accumulation at corners and switchbacks in the fuel cell flow channel [43].

322

Polymer Electrolyte Fuel Cell Degradation

possible and manage water content through thermal or other transport mechanisms to eliminate these effects and reduce water content. Dry purging should be done with careful attention to the purge gas temperature. An MK 513 series single cell of Ballard Power Systems Inc. [5,44] experienced freeze damage after a dry hot purge (dry N2 purge was conducted right after operation for one minute at 85 C on both sides). On the other hand, no damage occurred when the cell was cooled down to ambient temperature and then purged with cold dry N2. Although no reason behind this observation was disclosed, it’s important to note that the MK 513 series cell has very long, parallel flow channels. The hot purge may have over-dried the MEA near the inlet and then cooled and wetted the MEA near the outlets, leading to freezedamage. However, a cold purge induces less gradients of moisture leading to slower evaporation but less damage and a more uniform water distribution. The key point is removal of liquid water in contact with the catalyst layer without inducing damage from uneven stress caused by drying. Although water removal is necessary for freeze damage mitigation, it is very important not to over-dry the MEA. A dehydrated membrane will have a very low electric conductivity and cold start-up will not be possible. HFR measures the ionic resistance and therefore can be used as a diagnostic tool to determine optimal purge duration. HFR is not affected when the purging process removes residual liquid water from the channels and DM. Cell resistance starts to increase when water removal is initiated at the membrane level. An optimum strategy is to stop the purge at the inflection point of the resistance versus time curve [5] as show in Fig. 6.19 for an Mk9 10-cells stack. Vacuum purging was proposed as a method for drying out the DM and MEA of a cell [45], before shut-down in a freezing environment, since water is more easily drained at higher temperatures because of better evaporation. It’s

FIGURE 6.19 HFR change with purge time [5].

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

323

preferable to start vacuum purging as soon as the cell is shut down from its operating temperature. Vacuum drying at a higher temperature dictates the need for a smaller vacuum pump which is typically already onboard a vehicle for other purposes. Although this approach is shown to result in damage mitigation, and may be appropriate in certain niche applications, in general it is not generally believed to be practical in operating systems. Temperature-gradient driven water transport is an attractive non-parasitic water drainage method during fuel cell shut-down. The use of engineered temperature gradients within the stack has been demonstrated to prevent freeze damage [46,47]. There are two basic modes of temperature-gradient driven flux of water that are relevant at shut-down; thermo-osmotic transport in the membrane, and phase-change-induced (PCI) flux through the open voids. Thermo-osmosis in the membrane is the water flux observed when water with different temperatures is separated by the membrane [44,46,48–54]. Thermoosmotic water flux in fuel cell membranes is from the cold to the hot side, and depends on the difference in entropy between water stored in the membrane and water external to the membrane [53]. Unbound water transport is thermodynamically favored in the direction with increasing entropy [52,53]. Kim et al. further investigated water flux through the membrane and concluded that water flux is proportional to temperature difference as shown in Fig. 6.20, and inversely proportional to membrane thickness as seen in Fig. 6.21. An Arrhenius rate law was determined to capture this transport mode. PCI flow occurs with the presence of a temperature gradient and gas phase in the CL, MPL or main DM and dominates once irreducible saturation is

FIGURE 6.20

Thermo-osmotic water flux in Nafion 112 membrane [55].

324

Polymer Electrolyte Fuel Cell Degradation

FIGURE 6.21 Comparison of thermo-osmotic water flux of membranes [55].

attained in the porous media [54–56]. PCI flow is strongly dependent on average membrane temperature and temperature gradients [54–56]. The effect of DM/CL thermal mass was negligible. M. Khandelwal and Mench showed that thermo-osmotic flow can either assist or oppose PCI flow depending on the hydrophobic properties of the membrane [57]. In fuel cell media, it generally opposes the PCI flow. Thus, a residual water content in the warmer electrode can result under significant temperature gradients, which would tend to occur near the end plates of a stack. Therefore, to minimize water in the cathode CL, thermo-osmosis flux across the membrane is very important to help freeze durability. Both PCI flow and thermo-osmosis in various membranes and DM material sets have been experimentally investigated and quantified, and it was determined that both modes of transport can be wellcorrelated using Arrhenius rate laws as shown in Fig. 6.22. Although the type of reinforcement in the membrane has some impact, thermo-osmosis is fairly constant for perfluorosulfonic type membranes, but significantly less than regular concentration-based diffusion. Therefore, this mode of transport is not normally critical during operation, given the existing high range of uncertainty in published diffusivity values. However, during shut-down to a frozen state, thermo-osmosis can become important, as it can counteract the PCI flow, which moves liquid toward the cold location. The result of the interaction can be a residual frozen water saturation in the warmer-side catalyst layer of the MEA, as has been shown via recent modeling of this effect [57]. In general, the PCI flow is much more significant for even the small temperature gradients expected during shut-down between MEA components [56]. Several fuel cell manufacturers have also investigated this effect, and Ballard suggested

Chapter | 6

NR-MEA/SGL10BB Y = -1.486 - 1870X R-MEA/SGL10BB Y = -1.5974 - 1849.2X R-PEM A/SGL10BB Y = -1.6477 - 1858.9X Nafion 112/SGL10BB Y = -2.1236 - 1723.6X R-PEM A/SGL10AA Y= -4.767 - 998X

-6.5 Diffusivity, log10|D| (kg/m-K-s)

325

Freeze Damage to Polymer Electrolyte Fuel Cells

-7.0

Phase change induced flow (hot to cold)

-7.5

-8.0

T=10K T= 5K T= 3K

Thermo-osmosis (Cold to Hot)

-8.5 0.0026

0.0028

0.0030

0.0032

0.0034

0.0036

Reciprocal Temperature (1/K) FIGURE 6.22 Correlated thermo-osmosis and PCI flow relationships based on an Arrhenius rate law. Data from different temperature gradients all conveniently collapse into a single curve for a given material set [54,55].

a unique stack design to promote internal temperature gradients near end-plate locations to avoid damage and promote reliable start-up from a frozen condition [46].

4.1. Damage Mitigation via Material Choice and Design Although the various presented methods of mitigation are useful, in principle no action would be needed at shut-down if the operational overhead of liquid water was reduced to a value below that at which damage occurs. That is, if proper materials and design to reduce the liquid water overhead can be chosen, the required parasitic purge can be reduced. Work by Turhan et al. has shown that water content in a fuel cell can be reduced by as much as 50% with little performance change, just by changing the DM thickness and channel/land design [58]. The following parameters have been determined from accumulated research to be key controlling parameters in the water content in the fuel cell porous media and flow channels: 1. The operating conditions: It should be noted that high current does not necessarily result in higher water content. In fact, the high channel flow rates and heat produced by inefficiency often reduce water content as current increases. Low current conditions often have the greatest total stored water content in the fuel cell. 2. The thermal boundary conditions and heat transport: PCI flow plays a critical role in water distribution, as proven by various studies. Water

326

3.

4.

5.

6.

7.

Polymer Electrolyte Fuel Cell Degradation

distribution and storage can be controlled through manipulation of this boundary condition via coolant channel design or material selection. The material choices: Tremendous shifts in water content at similar operating conditions have been observed depending on the thickness and type of diffusion media and other components. The channel/land interface: The shape and surface energy (e.g. contact angle) have been shown to be critical in the drainage of liquid from accumulation under the lands, as described in [58]. This impact should not be overlooked in terms of expected water content and freeze effects. In general, liquid flow across this interface is dominated by capillary action, so that the interface shape, roughness, contour, and surface energy are important aspects of drainage. A hydrophilic interface is preferred to allow drainage from the DM into the channel. Manifold design: The ability for water to drain from the internal channel structure into the main manifold is a key factor. In this location, even a small amount of water can impede the ability to properly start-up from a frozen state. Thus, it is critical that this location remain free of accumulation at shut-down. Channel design: As described, there is a general desire to reduce the number of flow switchbacks and flow deceleration points to avoid water accumulation. Thus, the general design paradigm from this result is to straighten the flow path as much as possible, and maintain water balance through other means such as boundary temperature control. Channel shape and surface energy: It has been shown by many researchers in the fuel cell and micro-fluidics field, that there is a clear relationship between water retention in the channels and channel shape and surface energy. However, using hydrophobic channels is not a particularly good solution since it restricts water removal from the DM and can result in operational instabilities related to the creation of multiple slugs of water [59].

Many of these parameters are not included in modern computational models. Thus, there is still a tremendous discrepancy between the water distribution predicted and that which is observed in practice [60]. Clearly, much additional research is needed before this can be fully resolved.

4.2. Comments on Proper Conditions for Experimental Testing of Freeze/Thaw As discussed, there are discrepancies between the results in literature for seemingly similar testing. The issues which result in these discrepancies include: 1. Differences in the experimental configuration or materials. As discussed, the stiffness of the DM, membrane thickness and type, as well as channel to land width ratio and compression play a strong role in the development of freeze/thaw damage.

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

327

2. For single cells, it is imperative that precise thermal boundary conditions are maintained. Single-cell testing has traditionally taken place using a heating cartridge to maintain temperature, but neutron imaging has shown this to result in very different internal water distribution than if coolant channels are used. For accurate testing at the single cell level, it is imperative that coolant-based or other type boundary temperature control is used that is superior to cartridge heaters, which provide inconsistent and non-representative heating and cooling behavior. 3. The shut-down procedure used in the laboratory is obviously critical and should be carefully considered and controlled in terms of thermal boundary conditions. One of the main differences between single cell and stack cell freeze/thaw testing is that in a single cell, both sides are colder than the center of the cell at shut-down. This results in PCI flow removal of water from both electrodes, and mitigation of freeze damage compared to an instack cell, where the temperature gradient is in one direction on both sides of the membrane. In-stack cell testing can be simulated by using dual coolant controlled boundary conditions. If separate coolant flows are used, a temperature gradient representative of any particular location in the fuel cell stack can be simulated. 4. The initial conditions before shut-down and purge to a frozen state should also be carefully maintained. That is, the same shut-down procedure, executed on two similar cells with a different operational history, will result in a different final condition before freeze. This discrepancy can be eliminated by ending a cycle with a pre-shut-down step. By operating at a selected known condition for a significant period of time and then initiating shutdown, the previous operational history effects can be effectively erased, and the cells will be shut-down from a precise initial condition. For example, for a 50 cm2 active area single cell, operation at 0.6 V for 30 minutes before initiating the purge protocol should eliminate any differences that might arise from operational conditions before purge.

5. SUMMARY AND FUTURE OUTLOOK This chapter has examined the results of published studies that examine physicochemical degradation in polymer electrolyte fuel cells resulting from a shut-down to a frozen state. Damage caused from a frozen start-up is out of the scope of this publication, but deserves additional attention in the literature. Ultimately, no damage as a result of freezing to 40 C was observed for any common fuel cell materials if there was no contact with liquid water. This result indicates that damage-free shut-down to a frozen state is possible through proper engineering of operational and shut-down protocol, materials, and design. Achieving a damage-free frozen condition is difficult, however, because the time for purge should be short, and parasitic losses should be minimized.

328

Polymer Electrolyte Fuel Cell Degradation

Four to six years ago, the major automotive manufacturers reported successful start-up of their respective fuel cell vehicles in relation to the issue of freeze. It seems that their approach was through good systems engineering leading to a multitude of patents. In fact, sustaining a freezing environment is not the challenge holding up fuel cell vehicle commercialization. On the other hand, the 2007 DOE fuel cell technical plan reports that clear results of degradation rate over a 5,000 hour lifespan (150,000 miles equivalent) of an automotive stack have not been declared, although it is estimated to be < 20%. The ultimate goal is 5% performance degradation at the end of life of a stack subjected to the full range of external environmental conditions (40 to 40 C). From a summary of the existing literature, damage to the fuel cell components is a result of water expansion upon freezing, and a frost-heave delamination mechanism unrelated to the expansion process. Electrochemical surface area (ECSA) reduction has been commonly measured as a function of frozen conditions. Physical damage to PEFC components were identified to include membranejCL delamination, CLjDM delamination, and local pore damage in porous layer (CL and DM), and some membrane cracking. Loss of DM hydrophobicity and some morphological changes have also been observed, including some instances of DM punch-through from ice formation. A key source of freeze damage is now known to be the result of liquid water contact with the ionomer in the CL and membrane at shut-down. After a frozen condition is reached, the excess water uptake in the membrane can cause significant local delamination damage along the CL interface. Key factors which influence the degree of damage include the compression distribution on the MEA, membrane type and thickness, diffusion media stiffness, and shut-down conditions. Designs which limit areas of low compression are better suited for a frozen environment. Stiff diffusion media materials and thinner membranes with reinforced structure offer the greatest resistance to damage by limiting expansion and contraction forces, potential interfacial accumulations of water, and membrane-based sources of water under a frozen state. The ideal shut-down condition appears to be one in which the membrane phase is slightly and uniformly under-humidified, ensuring a low level of liquid accumulation and contact with the ionomer. Overly drying the membrane results in a poor cold start and potential membrane damage from internal stress generation. In order to achieve the desired shut-down condition of a slightly and uniformly under-humidified membrane, a simple, low temperature dry purge is effective, but too time consuming and parasitic to achieve desired levels of performance in practical operating systems. Various different purge approaches and damage mitigation techniques have been developed. Among the most promising is the use of controlled temperature gradients to assist liquid water

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

329

drainage via a phase change induced flow. Engineering design of typical stack components or coolant flow can induce sufficient gradients during shut-down to assist water removal from porous media into the channels, which can then be flushed by a short blast purge under low temperature conditions that will not harm the membrane. Many of the discrepancies between the experimentally observed phenomena can be attributed to the different materials, channel/land configurations and operational protocols. It is critical in freeze/thaw testing to achieve proper thermal boundary conditions and initial conditions for shut-down to a frozen state to assure reliable data and interpretation. A key difference between single cell and in-stack data is the thermal boundary conditions, which can control the final distribution of liquid going into a frozen state. A dual coolant system can be used to achieve near isothermal controlled boundary conditions, as well as to simulate accurate conditions for in-stack cells with a single laboratory cell. Although much work has been done to identify and explain freeze damage in PEFCs, there is still work to be done. The role of the CLjDM interface has been shown to be critical, yet little is known about the in-situ nature of this interface, particularly under dynamic operating conditions. Understanding the nature of materials and design so that the residual liquid water overhead in the fuel cell can be reduced before shut-down is perhaps more critical to achieve desired performance levels. If cells can be designed to have greatly reduced stored water content during operation, less is obviously required of the shutdown. Finally, a more complete knowledge of the nature of condensation and evaporation in fuel cell media is required for accurate modeling. Currently, models are constructed based on thermodynamic driving force of saturation pressure gradients and no information on the potentially important effects of surface energy or morphology are included.

ACRONYMS CL CV DM DSC ECSA F/T GDL HFR LANL MEA MPL PEFC PEM RH SEM

Catalyst Layer Cyclic Voltammetry Diffusion Media Differential Scanning Calorimetry Electrochemical Surface Area Freeze/Thaw Gas Diffusion Layer High Frequency Resistance Los Alamos National Lab Membrane Electrolyte Assembly Microporous Layer Polymer Electrolyte Fuel Cell Polymer Electrolyte Membrane Relative Humidity Scanning Electron Microscopy

330

Polymer Electrolyte Fuel Cell Degradation

REFERENCES [1] US Department of Energy, Hydrogen Fuel Cells and Infrastructure Technologies Program. (www1.eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/fuel_cells.pdf), 2005. [2] H. Liu, Dimensionally Stable High Performance Membrane, Proceedings of 2006 DOE Annual Program Review, DOE, Washington DC, 2006. [3] S. Kim, M.M. Mench, Physical Degradation of Membrane Electrode Assemblies Undergoing Freeze/Thaw Cycling: Micro-Structure Effects, J. Power Sources 174 (2007) 206–220. [4] S. Kim, B.K. Ahn, M.M. Mench, Physical Degradation of Membrane Electrode Assemblies Undergoing Freeze/Thaw Cycling: Diffusion Media Effects, J. Power Sources 179 (2008) 140–146. [5] J. St-Pierre, J. Roberts, K. Colbow, S. Campbell, A. Nelson, PEMFC Operational and Design Strategies for Sub Zero Environments, J. New Mater. Electrochem. Syst. 8 (2005) 163–176. [6] S. He, M.M. Mench, One-Dimensional Transient Model for Frost Heave in Polymer Electrolyte Fuel Cell – Part I: Physical Model, J. Electrochem. Soc. 153 (2006) A1724–A1731. [7] S. He, S.H. Kim, M.M. Mench, One-Dimensional Transient Model for Frost Heave in Polymer Electrolyte Fuel Cell – Part II: Parametric Study, J. Electrochem. Soc. 154 (2007) B1024–B1033. [8] S. He, J.H. Lee, M.M. Mench, One-Dimensional Transient Model for Frost Heave in Polymer Electrolyte Fuel Cell – Part III: Heat Transfer, Microporous Layer, and Cyclic Effect, J. Electrochem. Soc. 154 (2007) B1227–B1236. [9] M. Khandelwal, S. Lee, M.M. Mench, One-dimensional Thermal Model of Cold-Start in a Polymer Electrolyte Fuel Cell Stack, J. Power Sources 172 (2007) 816–830. [10] C. Hartnig, I. Manke, R. Kuhn, N. Kardjilov, J. Banhart, W. Lehnert, Cross–sectional insight in the water evolution and transport in polymer electrolyte fuel cells, Appl. Phys. Lett. 92 (2008) 1–3. 134106. [11] E.E. Kimball, J.B. Benziger, and Y.G. Kevrekidis, Effects of GDL Structure with an Efficient Approach to the Management of Liquid Water in PEM Fuel Cells, Fuel Cells, (2010) DOI: 10.1002/fuce.200900110. [12] T. Swamy, E.C. Kumbur, M.M. Mench, Characterization of Interfacial Structure in PEFCs: Water Storage and Contact Resistance Model, J. Electrochem. Soc. 157 (2010) B77–B85. [13] S. Kim, Ph.D. Thesis, The Pennsylvania State University, 2008. [14] Q. Yan, H. Toghiani, Y. Lee, K. Liang, H. Causey, Effects of Sub-Freezing Temperatures on a PEM Fuel Cell Performance, Startup and Fuel Cell Components, J.Power Sources 160 (2006) 1242–1250. [15] J.P. Meyers, Fundamental Issues in Subzero PEMFC Startup and Operation, Proceedings of Fuel Cell Operations at Sub–Freezing Temperature Workshop, DOE, Washington DC, (2005), (http://www1.eere.energy.gov/hydrogenandfuelcells/fc_freeze_workshop.html). [16] R. Gaylord, Stationary Application and Freeze/Thaw, Proceedings of Fuel Cell Operations at Sub-Freezing Temperature Workshop, DOE, Washington DC, (2005), (http://www1.eere. energy.gov/hydrogenandfuelcells/fc_freeze_workshop.html). [17] R.C. McDonald, C.K. Mittelsteadt, E.L. Thompson, Effects of Deep Temperature Cycling on NafionÒ 112 Membranes and Membrane Electrode Assemblies, Fuel Cells 4 (2004) 208–213. [18] Q. Guo, Z. Qi, Effect of Freeze-Thaw Cycles on the Properties and Performance of Membrane-Electrode Assemblies, J. Power Sources 160 (2006) 1269–1274. [19] J. Hou, H. Yu, S. Zhang, S. Sun, H. Wang, B. Yi, P. Ming, Analysis of PEMFC Freeze Degradation at 20 C After Gas Purging, J. Power Sources 162 (2006) 513–520.

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

331

[20] R. Mukundan, Y.S. Kim, F. Garzon, B. Pivovar, 2006 DOE Hydrogen Program Annual Progress Report (2006) 926–929. [21] E.A. Cho, J.J. Ko, H.Y. Ha, S.A. Hong, K.Y. Lee, T.W. Lim, I.H. Oh, Characteristics of the PEMFC Repetitively Brought to Temperatures Below 0 C, J. Electrochem. Soc. 150 (2003) A1667–A1670. [22] K. Tajiri, Y. Tabuchi, C.Y. Wang, Isothermal Cold Start of Polymer Electrolyte Fuel Cells, J. Electrochem. Soc. 154 (2007) B147–B152. [23] G. Xie, T. Okada, The State of Water in Nafion 117 of Various Cations Forms, Denki Kagakuoyobi Butsuri Kagaku 64 (1996) 718–726. [24] J.W. Paquette, K.J. Kim, Behavior of Ionic Polymer-metal Composites Under Subzero Temperature Conditions, Proc. ASME Intl. Expo, 205 ASME (2003). [25] J. Newman, Investigating Failure in Polymer-Electrolyte Fuel Cells, US DOE Annual Program Review Proc. (2005). [26] T.A. Zawodzinski, Membranes and MEAs at Freezing Temperatures, DOE Fuel Cell Operations at Sub-Freezing Temperatures Workshop, (2005). [27] M. Pineri, G. Gebel, R.J. Davies, O. Diat, Water Sorption-desorption in NafionÒ Membranes at Low Temperature, Probed by Micro X-ray Diffraction, J. Power Sources 172 (2007) 587–596. [28] H. Liu, Dimensionally Stable High Performance Membrane, Proceedings of 2008 DOE Annual Program Review, DOE, Washington DC, 2008. [29] R. Borup, J. Meyers, B. Pivovar, Y.S. Kim, R. Mukundan, N. Garland, D. Myers, M. Wilson, F. Garzon, D. Wood, P. Zelenay, K. More, K. Stroh, T. Zawodzinski, J. Boncella, J.E. McGrath, M. Inaba, K. Miyatake, M. Hori, K. Ota, Z. Ogumi, S. Miyata, A. Nishikata, Z. Siroma, Y. Uchimoto, K. Yasuda, K.-i. Kimijima, N. Iwashita, Scientific Aspects of Polymer Electrolyte Fuel Cell Durability and Degradation, Chem. Rev. 107 (2007) 3904–3951. [30] F.E. Hizir, S.O. Ural, E.C. Kumbur, M.M. Mench, Characterization of Interfacial Morphology in Polymer Electrolyte Fuel Cells: Micro-porous Layer and Catalyst Layer Surfaces, J. Power Sources 195 (2010) 3463–3471. [31] R. Alink, D. Gerteisen, M. Oszcipok, Degradation Effects in Polymer Electrolyte Membrane Fuel Cell Stacks by Sub-Zero Operation – An In Situ and Ex Situ Analysis, J. Power Sources 182 (2008) 175–187. [32] S. Ge, C.Y. Wang, Cyclic Voltammetry Study of Ice Formation in the PEFC Catalyst Layer during Cold Start, J. Electrochem. Soc. 154 (2007) B1399–B1406. [33] T. Hottinen, O. Himanen, S. Karvonen, I. Nitta, Inhomogeneous Compression of PEMFC Gas Diffusion Layer, J. Power Sources 171 (2007) 113. [34] V. Mishra, F. Yang, R. Pitchumani, Measurement and Prediction of Electrical Contact Resistance Between Gas Diffusion Layers and Bipolar Plate for Applications to PEM Fuel Cells, ASME J. Fuel Cell Sci. Technol. 1 (2004) 2. [35] U. Beuscher, Simon J.C. Cleghorn, W.B. Johnson, Challenges for PEM Fuel Cell Membranes, Intl. J. Energy Res. 29 (2005) 1103–1112. [36] A.K. Srouji, Cold Start Induced Damage to Polymer Electrolyte Fuel Cell, Masters Thesis, The Pennsylvania State University, 2010. [37] C. Chacko, R. Ramasamy, S. Kim, M. Khandelwal, M.M. Mench, Characteristic Behavior of Polymer Electrolyte Fuel Cell Resistance During Cold Start, J. Electrochem. Soc. 155 (2008) 1145–B1154. [38] A.A. Pesaran, G. Kim, G.D. Gonder, PEM Fuel Cell Freeze and Rapid Startup Investigation, Milestone Report, NREL/MP 540 (2005) 8760.

332

Polymer Electrolyte Fuel Cell Degradation

[39] M. Khandelwal, M.M. Mench, Direct Measurement of Through-plane Thermal Conductivity and Contact Resistance in Fuel Cell Materials, J. Power Sources 161 (2006) 1106–1115. [40] K.T. Cho, M.M. Mench, Fundamental Characterization of Evaporative Water Removal from Fuel Cell Diffusion Media, J. Power Sources 195 (2009) 3858–3869. [41] K.T. Cho, M.M. Mench, Coupled Effects of Flow Field Geometry and Diffusion Media Material Structure on Evaporative Water Removal from Polymer Electrolyte Fuel Cells, International Journal of Hydrogen Energy 35 (2010) 12329–12340. [42] K.T. Cho, M.M. Mench, Effect of Material Properties on Evaporative Water Removal from Polymer Electrolyte Fuel Cell Diffusion Media, J. Power Sources 195 (2010) 6748–6757. ¨ nlu¨, Study of [43] N. Pekula, K. Heller, P.A. Chuang, A. Turhan, M.M. Mench, J.S. Brenizer, K. U Water Distribution and Transport in a Polymer Electrolyte Fuel Cell using Neutron Imaging, Nucl. Instru Meth, in Phys. Res. Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 542 (2005) 134–141. [44] M. Perry, T. Patterson, J. ONeill, PEM Fuel Cell Freeze Durability and Cold Start Project, US DOE Annual Program Review Proceedings, DOE, Washington, DC, (2007). http://www.eere. energy.gov. [45] R.L. Fuss, Freeze-Protecting a Fuel Cell by Vacuum Drying, United States patent, 6358637 (2002). [46] R. Bradean, H. Haas, K. Eggen, C. Richards, T. Vrba, Stack Models and Designs for Improving Fuel Cell Startup from Freezing Temperatures, ECS Trans. 3 (2006) 1159–1168. [47] A.Z. Weber, J. Newman, Coupled Thermal and Water Management in Polymer Electrolyte Fuel Cells, J. Electrochem. Soc. 153 (2006) A2205. [48] M. Perry, T. Patterson, J. ONeill, Start and Operation of PEMFC Stacks under Sub-Freezing Conditions, proceedings of 5th International Conference on Fuel Cell Science Engineering and Technology, ASME (2007). [49] R. Zaffou, H.R. Kunz, J.M. Fenton, Temperature-Driven Water Transport in Polymer Electrolyte Fuel Cells, ECS Trans. 3 (2006) 909–913. [50] R. Zaffou, J.S. Yi, H.R. Kunz, J.M. Fenton, Temperature-Driven Water Transport through Membrane Electrode Assembly of Proton Exchange Membrane Fuel Cells, Electrochem. Solid-State Lett. 9 (2006) A418–A422. [51] J.P.G. Villaluenga, B. Seoane, V.M. Barragan, C. Ruiz-Bauza, Thermo-osmosis of Mixtures of Water and Methanol through a Nafion Membrane, J. Membr. Sci. 274 (2006) 116–122. [52] M. Tasaka, T. Hirai, R. Kiyono, T. Aki, Solvent Transport across Cation-Exchange Membranes under a Temperature Difference and under an Osmotic Pressure Difference, J. Membr. Sci. 71 (1992) 151–159. [53] M. Tasaka, T. Mizuta, O. Sekiguchi, Mass Transfer through Polymer Membranes due to a Temperature Gradient, J. Membr. Sci. 54 (1990) 191–204. [54] S. Kim, M.M. Mench, Temperature Gradient Induced Water Transport in Polymer Electrolyte Membranes, ECS Trans., Phoenix meeting (2008). [55] S. Kim, M.M. Mench, Investigation of Temperature-Driven Water Transport in Polymer Electrolyte Fuel Cell: Thermo-osmosis in Membranes, J. Memb. Sci. 328 (2008) 113–120. [56] M. Khandelwal, S. Lee, M.M. Mench, Model to Predict Temperature and Capillary Pressure Driven Water Transport in PEFCs After Shutdown, J. Electrochem. Soc. 156 (2009) B703–B715. [57] M. Kandelwal, M.M. Mench, J. Power Source (2010). in press. [58] A. Turhan, K. Heller, J.S. Brenizer, M.M. Mench, Passive Control of Liquid Water Storage and Distribution in a PEFC through a Flow-Field Design, J. Power Sources 180 (2008) 773–783.

Chapter | 6

Freeze Damage to Polymer Electrolyte Fuel Cells

333

[59] A. Turhan, S. Kim, M. Hatzell, M.M. Mench, Impact of Channel Wall Hydrophobicity on Through-plane Water Distribution and Flooding Behavior in a Polymer Electrolyte Fuel Cell, Electrochim. Acta 55 (2010) 2734–2745. [60] M.M. Mench, Advanced Modeling in Fuel Cell Systems: A Review of Modeling Approaches, Proceedings of the 18th World Hydrogen Energy Conference, 2010. [61] X. Yu, B. Zhou, A. Sobiesiak, Water and Thermal Management for Ballard PEM Fuel Cell Stack, J. Power Sources 147 (2005) 184–195. [62] M.S. Wilson, J.A. Valerio, S. Gottesfeld, Low Platinum Loading Electrodes for Polymer Electrolyte Fuel Cells Fabricated using Thermoplastic Ionomers, Electrochim. Acta 40 (1995) 355–363. [63] T. Patterson, R. Balliet, PEM Fuel Cell Freeze Durability and Cold Start Project, DOE Hydrogen Program Annual Progress Report (2006) 910–912. [64] T. Patterson, J.P. Meyers, PEM Fuel Cell Freeze Durability and Cold Start Project, Proceedings of 2006 DOE Annual Program Review, DOE, Washington DC, 2006. [65] E.A. Cho, J.J. Ko, H.Y. Ha, S.A. Hong, K.Y. Lee, T.W. Lim, I.H. Oh, Effects of Water Removal on the Performance Degradation of PEMFCs Repeatedly Brought to < 0 C, J. Electrochem. Soc. 151 (2004) A661–A665. [66] M. Oszcipok, M. Zedda, D. Riemann, D. Geckeler, Low Temperature Operation and Influence Parameters on the Cold Start Ability of Portable PEMFCs, J. Power Sources 154 (2006) 404–411. [67] M. Oszcipok, D. Riemann, U. Kronenwett, M. Kreideweis, M. Zedda, Statistic Analysis of Operational Influences on the Cold Start Behaviour of PEM Fuel Cells, J. Power Sources 145 (2005) 407–515. [68] R. Mukundan, Y.S. Kim, F. Garzon, B. Pivovar, Freeze/Thaw Effects in PEM Fuel Cells, ECS Trans. 1 (8) (2006) 403–413.